Advice, stories, and expertise about work life today.



We know your inbox is protected space, so we promise to send only the good stuff, twice a month.

Public Sector

What we can learn from AI when planning urban spaces

Artificial intelligence is being used to understand cities and build new types of structures – and it’s becoming clear that there’s a lot we can learn from it.

Siri, Alexa, and the other personal assistants that so many of us have in our homes, cars, and pockets are designed to learn from us. They familiarize themselves with facets of our lives – what routes we take to work, what musicians we like to listen to – so that they can perform more efficiently and improve a world of artificial intelligence-powered devices. But what can we learn from the AI in order to improve our world? Turns out there’s a lot.

Academics and businesses are already exploring the ways in which AI can help us better understand our cities, uncovering information that was previously too laborious to quantify, and producing design solutions that challenge what’s humanly possible. And with this knowledge, new opportunities for how we develop urban spaces emerge.


Myth busting

Urban planning has long been informed by our understanding of a city’s needs; for example, bolstering public transport and high-density housing to meet population growth. But what if some of these understandings are actually misconceptions?

That’s a question AI can answer, based on a four-year research project undertaken by MIT’s Media Lab. The team studied how neighborhoods in five different cities have changed by analyzing 1.6 million pairs of photos. Their method involved comparing scenes taken seven years apart – something that would be prohibitively time-consuming for humans, but was achievable with the help of a human-trained machine learning system. Through the data, the researchers were able to test common beliefs about cities. For instance, that high and low incomes correlate with positive and negative changes (they don’t). Or that housing prices are consistent with revitalization (they aren’t).

Similarly, urbanism-focused AI startup Topos has rethought the established borough boundaries in New York City with a model that’s based on algorithms, not tradition. By examining existing information, such as satellite images, topographic data, business concentration, and population density, and then grouping areas based on similarities, the company has suggested five modern boroughs that provide a more nuanced look at the city than the existing definitions allow.

Both of these approaches to investigating urban environments can provide governments and businesses with insights that are unavailable without the help of AI, providing a better understanding of changes within cities and allowing for more considered and effective urban planning.

Algorithmic design

Architects have long sought to push the boundaries of style, but the next frontier might be one that’s isn’t even explored by humans. Architect-programmers Michael Hansmeyer and Benjamin Dillenburger have spent years producing structures that are born from the mind of an AI instructed to work within set parameters. The results – such as the recent Digital Grotesque II, with its 1.35 billion surfaces – are overwhelmingly intricate and distinct from existing architecture. Hansmeyer estimates that manually designing something so detailed would take "thousands" of years.

And it isn’t just the artistic end of the spectrum that can benefit from autonomous methods. Various architects and startups are looking into how AI can enhance mass-produced homes. In his Mass Market Alternatives exhibition, architect John Szot has shown how algorithmic models can add some much-needed diversity to the uniform designs that dominate suburban residential developments. His program is intended to generate unique layouts based on existing dwelling styles that "challenge conventions and broaden our spatial and experiential palette."

Prefabricated housing company Cover, on the other hand, is letting owners play a part in the AI-powered planning. Whereas prefab properties tend to be homogenous to keep the design process short and inexpensive, Cover’s system is able to quickly and cheaply create personalized home arrangements by surveying customers on their lifestyles and needs, and considering environmental factors such as the sun’s path. Undoubtedly, AI presents opportunities for structural variations that exceed human capabilities and possibly imagination, and may even birth the next architectural trends.

As our urban spaces grow and change, AI looks set to become a valuable part of the planning process – not simply as a tool, but as an informative and creative force that can help us understand our world and provide new ways to approach design.

Social_Horizontal_Image_1080h x 1920w_S1_KNOWING-320554-edited

Urban Mobility

Optimize on-demand and scheduled mobility operations to enable seamless intermodal journeys.

Find out more

Sign up for our newsletter